Number Words & Number System

Laurent Sagart laurent.sagart at GMAIL.COM
Sat Apr 21 19:11:31 UTC 2007

Dear Richard,

I see what you are trying to do, connecting the Vanuatu forms for '5' with
xasep, the Pazeh form for 'five'. I am not competent to write about Oceanic
languages, so I will give my comments here for what they are worth.

Alex is right to point out that that your proposal requires sound
correspondences between the Formosan West coast word for 'five', which can
be reconstructed as PAN *RaCep, and you Vanuatu words. Like Alex, I suspect
this is not the case. Perhaps Oceanic specialists can tell us more.

The second point has to do with the shape of the Austronesian language tree.
In Taiwan, reflexes of *RaCep
are exclusively found on the west coast, and they are found nowhere else
outside of Taiwan either. The reason for this seems to be that after a
period of competition between *RaCep and *lima, *lima finally won out,
displacin Ra

2007/4/21, Richard Parker < richardparker01 at>:
> Many thanks, Alex, for responding so quickly. ( I recollect your name from
> the entry on Motlav in the ABVD, where you kindly added notes on your
> proto-Vanuatu (?) that helped me understand a lot).
> I'm no trained linguist (not knowing the difference between a morpheme and
> a phoneme gives the game away a bit), so what I am trying to do is not a
> comparative linguistics study, but to collect all the 'aberrant' number
> names and systems, (in 20% of all An languages) and see if there is any
> recognisable pattern.
> What I'm trying to do is to compare number systems and names (now that
> there is a lot of available material - thanks to Eugene Chan, who tells me
> he obtained the information from SIL ) over the whole Austronesian area,
> excepting the Western MP area that has mostly, but not wholly, applied the
> PAn symbolic number system. There are about 250 An languages east of the
> Wallace line which haven't done so, and some are very primitive indeed.
> >Yes, you are stretching things too much. Regular phonological
> correspondences should be looked at carefully before any comparison is
> proposed, especially at such a wide scale  (from Taiwan down to Vanuatu).
> Especially I would recommend having a look at the phonological
> reconstructions proposed for intermediate stages (POc, Proto North Central
> Vanuatu, etc.)
> I did try, very hard, at the beginning, to 'derive' the number-names from
> PAn. But when I got to trying to derive 'ha' for 4 (used either simply, or
> as a derivative)  in 110 languages (nearly 10% of all An languages), from
> PAn *Sepat, I just gave up, and started all over again, from scratch.
> >In the case of north Vanuatu (Banks islands) languages, which I know
> best, the variety of forms you point out can be shown, in each case, to boil
> down to just one proto-form (at the level of putative Proto-North Vanuatu,
> or perhaps Proto-North Central Vanuatu).
> I would certainly agree with that, and I thank you for pointing it out. My
> reason for quoting so many forms of basically the same morpheme was that
> Vanuatu languages use the same basic number morpheme more than any other
> particular group (and because Vanuatu has so many different languages that
> they take up a very prominent space in my chart).
> >For example, the set of forms "yeBe,leB, liBi, LaBea, livi"  all reflect
> regularly a form *lavea- 'formative for numbers between 6 to 9';  the forms
> with /r/ you cite (raB, rap, raf, raBe-) seem to reflect a proto-form
> *ravea- for languages further south.
> I don't know enough about phonological progression rules to know whether L
> becomes R or vice versa. I did think I was following some kind of
> progression when I saw 'du-a' = 2, and then 'lu-a' and 'ru-a' further east,
> but I came across Nelemwa (New Caledonia) where they seem to use du, lu, ru,
> all together, quite indiscriminately.
> >"six" is (Proto-North V) *lavea-te(w)a, from *lavea- + *tewa 'one';
> "seven" is *lavea-rua, from *lavea- + *rua 'two';
> "eight" is *lavea-tolu, from *lavea- + *tolu 'three';
> "nine" is *lavea-vati, from *lavea- + *vati 'four'.
> That system of number construction is very common all across the area from
> Flores to Vanuatu. Manus Island is a notable exception - many languages
> there go from 5 to 6 to 7=3x, 8=2x, 9=1x, then 10. A few very primitive
> number systems go from 5 to 5-1, 5-2, 4-4, 1X to 10 - (they're usually the
> same languages that also go 1, 2, 2-1, 2-2, 5).
> >There is no known etymology for this form. Semantically one would expect
> a connection with 'five', POc *lima  (through **l(im)avea??) but this is far
> from clear.
> I think I might have come across one, with the morpheme 'rabu' from Dawawa
> (2) and 'labui'= 2 when used to make the number 7 in  Igora, Wagawaga,
> Bohutu, and Suau - all of these are from Eastern New Guinea.
> >For another thing, "five" in this northern Vanuatu area comes from a
> proto-form *tave-lima, obviously from POc *lima "hand; five". The origin of
> the first element *tave- is unknown.
> I was (sort of) hinting at something like that, by mentioning the Banks
> Island languages that prefix -lima with a morpheme like tave in 5. Nearly
> 200 of the 'aberrant' number systems that I'm looking at use te, ta, or a
> derivative of that, for number 1  (including your tuwale).
> >The form for "one" was initially *te(w)a , as appears in "six"; this has
> become an indefinite or partitive marker in several Torres and Banks
> languages (and hence a marker for negation, etc.). Used alone, the form for
> "one" was replaced at some stage by a form which can be reconstructed (still
> for this group of languages) as *tuwa[l]e.
> >Several of these forms have been discussed in the following article:
> >François, Alexandre. 2005. Unraveling the history of vowels in seventeen
> north Vanuatu languages. Oceanic Linguistics 44-2 (Dec 2005): 443-504.
> Honolulu: University of Hawai'i Press.
> I would very much appreciate a copy of your paper.
> >There are other references around, concerning the many languages you
> mentioned, as well as concerning reconstructions for numbers.
>  >There doesn't appear to be any likely connection between any of these
> forms, and Pazeh xaseb.
> Vanuatu (Port Vila) is 2500 km from East New Guinea (Huon Gulf). Taiwan is
> 3000km from West New Guinea (Cenderawasih/Geelvink Bay). All of the most
> varied and most 'primitive' An number systems I am looking at come from
> between those two ends of New Guinea. Maybe it's time to look again.
> I'm certainly no trained linguist, and I'm probably a fully-certified
> 'island crank' , but I would ask you to refer to this before passing final
> judgement:
> -  Laurent Sagart's recent paper, connecting the Tai-Kadai language family
> to Austronesian, and speculating on just three number names from Pazeh
> (Taiwan)
> regards
> Richard
> _______________________________________________
> An-lang mailing list
> An-lang at

Laurent Sagart
312 Thurston Avenue, Apt A31
Ithaca, 14850 NY

Tel.: +1-646-288-8757

Télécharger mes articles/Download my papers:
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>
-------------- next part --------------
An-lang mailing list
An-lang at

More information about the An-lang mailing list